blake at NXS.NET
Wed Jun 3 05:04:07 UTC 2009
This is quite similar to experiences we have had with them. Again the only carrier we have dropped for technical reasons.
> -----Original Message-----
> From: Jo Rhett [mailto:jrhett at netconsonance.com]
> Sent: Tuesday, June 02, 2009 9:59 PM
> To: David Hubbard
> Cc: nanog at nanog.org
> Subject: Re: Savvis quality?
> On May 27, 2009, at 10:35 AM, David Hubbard wrote:
> > Just wondering if anyone can tell me their
> > opinion on Savvis bandwidth/company preferably
> > from a web host perspective. Considering a
> > connection.
> I wouldn't touch them with a 10g pole. They were the first and only
> provider we have dropped for inability to provide reasonable service.
> 1. They have problems in the bay area (and I've heard other places but
> I can't confirm) coming up with ports to connect to people on. We had
> long since outgrown 100mb (was 1g or higher with everyone else) but
> they couldn't come up with a 1g port to sell us. Then when one became
> free, they demanded a 700mb commit to get it. After I argued that we
> never run ports at that level of congestion they backed down to a
> 500mb commit but that was as low as they'd go. They had no budget to
> deploy more ports in any of the bay area peering facilities.
> 2. Their national NOC staff was gut-stripped down to 3 people. 24
> hours a day I'd find the same person answering issues we reported.
> Often outages weren't resolved until they could wake the engineer up.
> (this isn't surprising in a small company, it's very surprising in a
> network the size of Savvis)
> 3. We had repeated issues that needed escalation to our salesperson
> for credit. We never got calls back on any of these, even when we had
> escalated through phone, email and paper letters to him.
> 4. One day they changed the implementation of their community strings
> to start putting other providers and international customers in their
> US-Customer-Only community strings. We escalated this issue through
> management, and the final conclusion was that their community strings
> advertised to us had to be inconsistent to meet their billing needs.
> (ie get peers to send them traffic they shouldn't have gotten) We
> were forced to drop using their community strings and instead build a
> large complex route-map to determine which traffic should be routed to
> them. That's nonsense, and was the final straw.
> In one of the marathon phone calls with the NOC staff about this, a
> NOC manager frankly told me that Savvis had been stripped and reamed,
> and they were just trying to stay alive long enough to sell the low-
> cost carcass to another provider.
> Yeah, I think that pretty much sums it up.
> Jo Rhett
> Net Consonance : consonant endings by net philanthropy, open source
> and other randomness
More information about the NANOG